Research helps fighter pilots communicate better
With the help of research, fighter pilots can be better at communicating with each other and with aerial combat control in threatening situations. FOI has investigated how flight training and various types of education can be combined to increase the effectiveness of how pilots work together.

To analyse the efforts of individual fighter pilots or the group, FOI has created two visualisations that allow an overview of the pilots’ firing opportunities and threat picture. Photo: Björn Tesch/FOI.
FOI, in Kista, is the location of the Air Force combat simulation centre, FLSC, where pilots can practice combat technique and tactics in a protected and secret environment. This is where FOI, in an assignment from the Armed Forces, has investigated how flight training and education can be more effective. At the same time as the pilots have been training, researchers have studied how the pilots can do so more effectively, as well as both how to measure it and to support the pilots in their training.
“Every press of a button and all target searches and oral communications are registered in a large database. By using various tools and methods that we have produced, we can then analyse complex scenarios and measure the pilots’ performance. The result becomes a support for the pilots and the training leader when they reflect on how the training went,” says Sinna Lindquist, researcher in the Defence & Security, Systems and Technology Division at FOI.
To analyse the efforts of individual fighter pilots or the group, FOI has created two visualisations that allow an overview of the pilots’ firing opportunities and threat picture. The researchers have also successfully investigated how machine learning can be used to analyse all oral communication sent during a flight session in the simulator. This is in part to judge the effectiveness of the communications, in part to discover eventual shortcomings in the conversations between pilots and aerial combat controllers. Work is now in progress to implement selected parts of the analytical methods.
“We use the models, which are practiced using machine learning, for automatic transcription of sound files to text. Then the communication is analysed and visualised in various ways to support the pilots and aerial combat controllers when they assess the training session,” explains Hanna Lilja, Research Engineer at FOI.
The language the pilots use is prescribed in a manual that describes the terminology that should be implemented and how different things should be expressed on the radio. The communications usually deal with the manoeuvres the pilots are carrying out or planning. The pilots also receive information – warnings, for example – from the aerial combat controller. In threatening situations where warnings are frequent, it is important that the pilots avoid filling the radio channel with other talk.
“The matter of priorities is interesting to look at; whether the pilots keep to the order of priorities. But to do this, a computer must determine what type of expression it is; so, we have trained a model to do this classifying,” says Hanna Lilja.
Part of the project has been conducted in cooperation with the U.S. Air Force Research Laboratory.